Hypoxia-inducible Factor-1α (HIF-1α) Promotes Cap-dependent Translation of Selective mRNAs through Up-regulating Initiation Factor eIF4E1 in Breast Cancer Cells under Hypoxia Conditions*

Background: Hypoxia promotes tumor growth, but connections to translation mechanisms are obscure. Results: Hypoxia-enhanced tumorsphere growth of breast cancer cells is HIF-1α-dependent, and HIF-1α up-regulates eIF4E1 in hypoxic cancer cells. Conclusion: HIF-1α promotes cap-dependent translation of selective mRNAs through up-regulating eIF4E1 in cancer cells at hypoxia. Significance: Our study provides new insights into the translation mechanisms in cancer cells under low O2 concentrations. Hypoxia promotes tumor evolution and metastasis, and hypoxia-inducible factor-1α (HIF-1α) is a key regulator of hypoxia-related cellular processes in cancer. The eIF4E translation initiation factors, eIF4E1, eIF4E2, and eIF4E3, are essential for translation initiation. However, whether and how HIF-1α affects cap-dependent translation through eIF4Es in hypoxic cancer cells has been unknown. Here, we report that HIF-1α promoted cap-dependent translation of selective mRNAs through up-regulation of eIF4E1 in hypoxic breast cancer cells. Hypoxia-promoted breast cancer tumorsphere growth was HIF-1α-dependent. We found that eIF4E1, not eIF4E2 or eIF4E3, is the dominant eIF4E family member in breast cancer cells under both normoxia and hypoxia conditions. eIF4E3 expression was largely sequestered in breast cancer cells at normoxia and hypoxia. Hypoxia up-regulated the expression of eIF4E1 and eIF4E2, but only eIF4E1 expression was HIF-1α-dependent. In hypoxic cancer cells, HIF-1α-up-regulated eIF4E1 enhanced cap-dependent translation of a subset of mRNAs encoding proteins important for breast cancer cell mammosphere growth. In searching for correlations, we discovered that human eIF4E1 promoter harbors multiple potential hypoxia response elements. Furthermore, using chromatin immunoprecipitation (ChIP) and luciferase and point mutation assays, we found that HIF-1α utilized hypoxia response elements in the human eIF4E1 proximal promoter region to activate eIF4E1 expression. Our study suggests that HIF-1α promotes cap-dependent translation of selective mRNAs through up-regulating eIF4E1, which contributes to tumorsphere growth of breast cancer cells at hypoxia. The data shown provide new insights into protein synthesis mechanisms in cancer cells at low oxygen levels.

Hypoxia promotes tumor evolution and metastasis, and hypoxia-inducible factor-1␣ (HIF-1␣) is a key regulator of hypoxia-related cellular processes in cancer. The eIF4E translation initiation factors, eIF4E1, eIF4E2, and eIF4E3, are essential for translation initiation. However, whether and how HIF-1␣ affects cap-dependent translation through eIF4Es in hypoxic cancer cells has been unknown. Here, we report that HIF-1␣ promoted cap-dependent translation of selective mRNAs through up-regulation of eIF4E1 in hypoxic breast cancer cells. Hypoxia-promoted breast cancer tumorsphere growth was HIF-1␣-dependent. We found that eIF4E1, not eIF4E2 or eIF4E3, is the dominant eIF4E family member in breast cancer cells under both normoxia and hypoxia conditions. eIF4E3 expression was largely sequestered in breast cancer cells at normoxia and hypoxia. Hypoxia up-regulated the expression of eIF4E1 and eIF4E2, but only eIF4E1 expression was HIF-1␣-dependent. In hypoxic cancer cells, HIF-1␣-up-regulated eIF4E1 enhanced cap-dependent translation of a subset of mRNAs encoding proteins important for breast cancer cell mammosphere growth. In searching for correlations, we discovered that human eIF4E1 promoter harbors multiple potential hypoxia response elements. Furthermore, using chromatin immunoprecipitation (ChIP) and luciferase and point mutation assays, we found that HIF-1␣ utilized hypoxia response elements in the human eIF4E1 proximal promoter region to activate eIF4E1 expression. Our study suggests that HIF-1␣ promotes cap-dependent translation of selective mRNAs through up-regulating eIF4E1, which contributes to tumorsphere growth of breast cancer cells at hypoxia. The data shown provide new insights into protein synthesis mechanisms in cancer cells at low oxygen levels.
Although global translation is suppressed at hypoxia, which contributes to conserve energy and to sustain survival during the period of inefficient ATP production (1-3), a subset of selective mRNAs believed to be involved in the adaptive responses to hypoxia are preferentially translated in cancer cells (4 -6). For example, high levels of the c-Myc and Cyclin-D1 oncoproteins, the proangiogenic factors VEGF and Tie2, and the translation initiation factor eukaryotic initiation factor 4E1 (eIF4E1) have been observed in many types of solid tumors (7)(8)(9)(10). In addition, eIF4G, ATF4, and ATF6 have been shown to remain associated with high molecular weight polysomes during anoxic stress (0% O 2 ) (11,12).
Several possible mechanisms for bypassing translation inhibition in mammalian cells at hypoxia (1% O 2 ) have been investigated: 1) uncoupling of oxygen-responsive signaling pathways from mTOR functions in breast cancer cells (1), 2) activation of initiation through an HIF-2␣⅐RBM4⅐eIF4E2 complex in glioblastoma cells (here the HIF-2␣⅐RBM4⅐eIF4E2 complex cap-tures the 5Ј-cap and targets mRNAs to polysomes for active translation and therefore evades hypoxia-induced translation repression (18)), 3) internal ribosome entry site (IRES)-dependent initiation in normal and cancerous cells (here the 5Ј-UTR regions of some oncogenic mRNAs harbor IRES structures, which facilitate direct ribosome binding independent of formation of eIF4F complex at the cap (19,20)), and 4) IRESindependent initiation of selective mRNAs in normal and cancerous cells (here Cyclin-dependent kinase inhibitor p75 Kip2 and VEGF mRNAs are selectively translated by an IRES-independent mechanism in normal and cancer cells (21)). Although these possible bypass pathways for regulating translation in hypoxic cancer cells have been studied, the translation mechanisms underlying the adaptive and malignant phenotype of tumors at hypoxia have remained obscure.
In this study, we investigated the roles of HIF-1␣ in translation of selective mRNAs in hypoxic breast cancer cells. We observed that hypoxia promoted cell proliferation and tumorsphere growth of breast cancer cells, but this promotion effect was HIF-1␣-dependent. In cancer cells, eIF4E1 was the dominant factor of the eIF4E family under both normoxia and hypoxia conditions. Although hypoxia significantly elevated the expression of eIF4E1 and eIF4E2, the level of cellular eIF4E3 was very low in breast cancer cells at normoxia and hypoxia. On the other hand, HIF-1␣ significantly up-regulated the expression of eIF4E1 but not that of eIF4E2. We observed that hypoxic cancer cells were more sensitive to the eIF4E-eIF4G interaction inhibitor 4EGI-1 compared with normoxic cancer cells, which suggests a key role of eIF4F-controlled translation initiation in hypoxic cancers. Consistently, we found that HIF-1␣ utilized hypoxia response elements (HREs) in the proximal promoter region of eIF4E1 to promote eIF4E1 expression. Our study revealed the hypoxia-dependent roles of eIF4E factors in breast cancer cells as mediated by HIF-1␣.
Hypoxia Treatment-Hypoxia treatment (1% O 2 ) was performed in a hypoxia incubator chamber (Stem Cell Technology Inc.) by supplying it with 1% O 2 (balanced by 5% CO 2 and 94% N 2 ) for 20 -25 min at a rate of over 10 liter/min with a pressure of 1.3-1.5 p.s.i. to get rid of trace of O 2 in the chamber. To minimize glucose depletion, we replaced the medium with fresh MEGM (4 ml/well in 6-well plates) every 24 h.
Cytoplasmic Extract/mRNA Preparation-Cytoplasmic VEGF-A 165 protein/mRNA preparation was performed with NE-PER Nuclear and Cytoplasmic Extraction Reagents following the manufacturer's instructions. About 2 ϫ 10 6 HMLER (CD44 high /CD24 low ) cells treated with DMSO (20 M), inactive analog 4EGI-N (20 M), or (E)-4EGI-1 in the indicated concentrations at hypoxia (1% O 2 ) overnight were harvested by trypsin and washed with cold 1ϫ PBS twice followed by centrifugation at 500 ϫ g for 2 min. After removing the supernatant, 200 l of ice-cold Cytoplasmic Extraction Reagent I (with protease inhibitor and phosphoprotease inhibitor) was added followed by vigorous vortexing for 15 s and incubation on ice for 10 min. 11 l of ice-cold Cytoplasmic Extraction Reagent II was added followed by vigorous vortexing for 5 s and incubation for 2 min. After vigorous vortexing for 5 s and centrifugation at 16,000 ϫ g for 5 min, the supernatant (cytoplasmic extract) was immediately transferred into a prechilled tube. Either the cytoplasmic extract was used for cytoplasmic VEGF-A 165 ELISA test with VEGF-A 165 ELISA kit, or mRNA purification performed using a TRIzol (Invitrogen)-chloroform-isopropyl alcohol-ethanol method followed by real time PCR with VEGF-A 165 -specific primers.
HIF-1␣ Knockdown Assays-Lentiviral particles harboring shRNA (human) HIF-1␣ (sc-35561-V) were ordered from Santa Cruz Biotechnology for HIF-1␣ knockdown assays. Procedures and reagents for control lentivirus production and lentivirus infection were adapted from the Broad RNAi Consortium protocols as described previously (57). Control and shRNA (h) HIF-1␣ lentivirus-infected HMLER (CD44 high / CD24 low ) cells used for mammosphere growth analyses were cultured for 3 days at normoxia and hypoxia. For the qPCR and Western blot assays, control and shRNA (h) HIF-1␣ lentivirus-infected HMLER (CD44 high /CD24 low ) cells were cultured for 24 h at normoxia and hypoxia. Total RNAs were extracted using TRIzol reagent (Qiagen), and cellular proteins were extracted with the above mentioned radioimmune precipitation assay buffer.
Transfection and Luciferase Assay-For the luciferase assay, the Ϫ171 to ϩ34 bp DNA fragment of human eIF4E1, which harbors the region 4, was cloned into SacI and NheI sites of pGL3-Basic vector with primers 5Ј-GGTGGGGGAGAGACT-GAGCTCCCCAGAAGCCTCTCGTTACTCACGCAGCC-3Ј (sense) and 5Ј-GCCAAAGGCGCTAGCCACCGGTTCGAC-AGTCGCCATCTTAGATCGATCTGATCGCACAACCGC-TCC-3Ј (antisense). The Ϫ65 to ϩ34 bp DNA fragment of human eIF4E was cloned into SacI and NheI sites of pGL3-Basic vector as a control construct with primers 5Ј-GGACATATC-CGTCACGTGGCGAGCTCCTGGCCAATCCGGTTTGAA-TCTC-3Ј (sense) and 5Ј-GCCAAAGGCGCTAGCCACCGG-TTCGACAGTCGCCATCTTAGATCGATCTGATCGCAC-AACCGCTCC-3Ј (antisense). Point mutation (ACGTG to AAAAG) was performed with QuikChange II site-directed mutagenesis kits (Agilent). The indicated constructs or construct combination (with HA-HIF1␣-pcDNA3 plasmid) was transfected into breast cancer cells HMLER (CD44 high /CD24 low ) and MDA-MB-231 with Lipofectamine TM LTX and PLUS TM reagent (Invitrogen) using a standard protocol. After 2 days of recovery after transfection, cells were cultured at normoxia or hypoxia (1% O 2 for 24 h). Luciferase activity was measured using the luciferase assay system (Promega) with a Top Count microplate scintillation counter (Canberra). Three independent experiments were performed, and statistical data (mean Ϯ S.D.) are shown.

Hypoxia Promotes Mammosphere Growth of Breast Cancer
Cells-The capacity of mammosphere (tumorsphere) formation and growth of single cancer cells is an important characteristic of tumorigenicity (55,58,59). To minimize the potential suppression effects caused by glucose deprivation on tumorsphere growth at hypoxia, we cultured HMLER (CD44 high / CD24 low ) breast cancer cells with double MEGM (4 ml/well in 6-well plate) at hypoxia and replaced the medium with fresh MEGM every 24 h. We cultured 1 ϫ 10 4 cells/well at normoxia and hypoxia for 4 days and counted the numbers of smaller (30 -40-m) and larger (40 -50-m) mammospheres/1000 cells every day (Fig. 1). At day 4, the small mammosphere count per 1000 cells was 113.6 Ϯ 7.3 at normoxia and 149.3 Ϯ 5.6 at hypoxia (1% O 2 ), and for large mammospheres, the numbers were 17.6 Ϯ 1.53 at normoxia and 24 Ϯ 1 at hypoxia (1% O 2 ). Thus, hypoxia (1% O 2 ) significantly promoted tumorsphere growth of breast cancer cells both in number and size compared with normoxia (20% O 2 ; Fig. 1, A and B). The total cell numbers at day 4 increased at hypoxia in comparison with those at normoxia (data not shown). Our data suggest that hypoxia (1% O 2 ) promotes cell proliferation and tumorsphere growth of these breast cancer cells.
Hypoxia-promoted Breast Cancer Cell Mammosphere Growth Is HIF-1␣-dependent-To examine whether HIF-1␣ plays a role in hypoxia-promoted tumorsphere growth, we cultured HMLER (CD44 high /CD24 low ) cells with and without transient knockdown of HIF-1␣ at normoxia and hypoxia conditions for 3 days. At day 3, we measured the mammosphere numbers and found that knockdown of HIF-1␣ largely reduced the mammosphere numbers at hypoxia but had no effect at normoxia for both classes of mammospheres (Fig. 2, A and B). These observations suggest that hypoxia-promoted cell proliferation and mammosphere growth of breast cancer cell are HIF-1␣-dependent.
eIF4E1 Is the Dominantly Expressed Member of the Three eIF4E Proteins in Breast Cancer Cells under Normoxia and Hypoxia Conditions-To investigate the expression of the three members of the eIF4E family in breast cancer cells at both nor-moxia (20% O 2 ) and hypoxia (1% O 2 for 24 h), we measured the cellular mRNAs of eIF4E1, eIF4E2, and eIF4E3 by qPCR. Under normoxia conditions, we found that eIF4E2 and eIF4E3 mRNAs are expressed only at 25 Ϯ 2.11 and 1.12 Ϯ 0.77% of eIF4E1 mRNAs, respectively. At hypoxia, cellular eIF4E2 and eIF4E3 mRNAs levels are only 10.8 Ϯ 2.32 and 0.37 Ϯ 0.56% of eIF4E1 mRNAs, respectively (Fig. 3, A and B). These data indicate that eIF4E1 is the dominantly expressed member of the eIF4E family in breast cancer cells at both normoxia and hypoxia. In comparison, the expression of eIF4E3 was very low (only 1% of eIF4E1 at normoxia and 0.3% at hypoxia).
Hypoxia-up-regulated Expressions of eIF4E1 and eIF4E3 Are HIF-1␣-dependent-To examine the roles of hypoxia and HIF-1␣ on eIF4E1, eIF4E2, and eIF4E3 transcription in hypoxic cancer cells, we cultured HMLER (CD44 high /CD24 low ) cells with and without HIF-1␣ transient knockdown at normoxia and hypoxia conditions for 24 h and measured cellular mRNAs by qPCR. We detected that hypoxia strikingly elevated the expression of eIF4E1 (increased to about 4-fold) and eIF4E2 (increased to about 1.8-fold) in comparison with that at normoxia (Fig. 3, C-E), indicating that hypoxia strongly up-regulates eIF4E1, moderately increases eIF4E2, and has only a small effect on eIF4E3 expression. These data are consistent with the aforementioned observations that the relative expression levels of eIF4E2 compared with eIF4E1 were decreased at hypoxia, which is caused by the fact that the increase of eIF4E1 (4-fold) is greater than that of eIF4E2 (about 1.8-fold). Interestingly, we detected that shRNA-HIF-1␣ abrogated hypoxia-promoted expression of eIF4E1 (Fig. 3, C-E) but did not significantly affect eIF4E2 expression at normoxia, indicating that hypoxiaup-regulated expression of eIF4E1 is HIF-1␣-dependent.
HIF-1␣-promoted eIF4E1 Expression Is Important for Capdependent Translation of a Subset of mRNAs-To further examine the roles of hypoxia and HIF-1␣ on eIF4E1, eIF4E2, and eIF4E3, we performed Western blot assays. We observed that hypoxia increased cellular eIF4E1 (about 3-fold) and eIF4E2 (about 1.2-fold) compared with that at normoxia. HIF-1␣ knockdown significantly decreased eIF4E1 (to about 0.3-fold), but not eIF4E2, in comparison with that at normoxia (Fig. 4A), indicating that hypoxia-promoted eIF4E1 expression is HIF-1␣-dependent. We observed that the cellular levels of eIF4E1 proteins were much higher than those of eIF4E2 and eIF4E3 proteins at both normoxia and hypoxia conditions. In particular, eIF4E3 was hardly detectable both at normoxia and hypoxia, indicating that its expression is largely sequestered in cancer cells at both normoxia and hypoxia.
To examine whether HIF-1␣ plays a role in translation of a subset of mRNAs, we tested cellular c-Myc, Cyclin-D1, eIF4G1, eIF1A, eIF5, and GAPDH with and without HIF-1␣ at normoxia and hypoxia. We observed that hypoxia elevated the cellular c-Myc, Cyclin-D1, and eIF4G1 proteins and that HIF-1␣ knockdown dramatically decreased the levels of these proteins but did not significantly reduce the levels of eIF1A, eIF5, or GAPDH (Fig. 4B), suggesting that HIF-1␣ plays a role in the protein expression of selective genes. Next, we observed that HIF-1␣ transient knockdown in hypoxic breast cancer cells did  not significantly decrease the cellular levels of c-Myc, Cyclin-D1, or eIF4G1 mRNAs (Fig. 4C). In addition, we found that there is no potential HRE (to which HIF-1␣ binds) in the promoter regions (Ϫ700 bp) of these three genes. These observations indicate that HIF-1␣-up-regulated c-Myc, Cyclin-D1, and eIF4G1 levels in hypoxic cancer cells are not likely caused by HIF-1␣-promoted transcription of these genes but by the roles of HIF-1␣ in translation. To investigate whether HIF-1␣-upregulated eIF4E1 mediates the expression of c-Myc, Cyclin-D1, and eIF4G1, we tested the efficacy of (E)-4EGI-1, an active 4EGI-1 isomer that binds eIF4E1 and inhibits the eIF4E1-eIF4G1 interaction (60), on these proteins in hypoxic breast cancer cells. We observed that (E)-4EGI-1 significantly inhibited hypoxia-elevated cellular c-Myc, Cyclin-D1, and eIF4G1 levels in breast cancer cells (Fig. 4D) but did not significantly affect the cellular levels of c-Myc, Cyclin-D1, and eIF4G1 mRNAs (data not shown), indicating that HIF-1␣-up-regulated eIF4E1 facilitates translation of c-Myc, Cyclin-D1, and eIF4G1 mRNAs.
VEGF, which is up-regulated by HIF-1␣ in hypoxic cancer cells (61), promotes cancer cell proliferation and tumor growth (62). We examined the effects of (E)-4EGI-1 on the expression of VEGF-A 165 in hypoxic breast cancer cells. We observed that (E)-4EGI-1 dramatically decreased hypoxia-promoted cytoplasmic VEGF-A 165 levels ( Fig. 4E) but did not significantly affect cytoplasmic VEGF-A 165 mRNA under the same conditions (Fig. 4F). These results suggest that hypoxia-promoted VEGF-A 165 translation in hypoxic breast cancer cells is largely regulated by eIF4E1-mediated cap-dependent translation. Taken together, the above data demonstrate that HIF-1␣-promoted eIF4E1 enhances cap-dependent translation of a subset of mRNAs in breast cancer cells under low oxygen conditions. c-Myc and Cyclin-D1 Are Involved in Breast Cancer Cell Mammosphere Growth at Hypoxia-To examine whether c-Myc and Cyclin-D1 are implicated in breast cancer cell mammosphere growth at hypoxia, we tested the efficacy of c-Myc inhibitor 10058-F4 (63) and Cyclin-D1 inhibitor PD 0332991 (64) on HMLER (CD44 high /CD24 low ) cell mammospheres. We observed that both inhibitors significantly inhibited hypoxiapromoted mammosphere growth of these breast cancer cells (Fig. 5, A and B), indicating that c-Myc and Cyclin-D1 are important for cancer cell mammosphere growth under hypoxia conditions.
To further investigate whether eIF4E1-mediated cap-dependent translation dominates in hypoxic cancer cells, we treated tumorspheres with (E)-4EGI-1 together with vehicle and the inactive analogue 4EGI-N at hypoxia and normoxia. As expected, we observed that (E)-4EGI-1 preferentially inhibited hypoxia-promoted tumorsphere growth, whereas vehicle-and 4EGI-N (40 M)-treated tumorspheres increased in both number and size (Fig. 5C). These results suggest that eIF4E1-regulated cap-dependent translation dominates (at least is indispensible) in mammosphere growth of breast cancer cells under hypoxia conditions. Together, these data demonstrate that HIF-1␣-promoted cap-dependent translation of selective mRNAs through up-regulating eIF4E1 contributes to breast cancer cell mammosphere growth at low oxygen levels.
HIF-1␣ Utilizes HREs in the Proximal Promoter of eIF4E1 to Promote Its Transcription in Hypoxic Cancer Cells-As eIF4E1 is the dominant member of the eIF4E family in breast cancer cells and is the key factor for cap-dependent translation initiation, here we focused on understanding the mechanism of how HIF-1␣ promotes eIF4E1 expression. We analyzed the promoter regions of human eIF4E1 genes. We found that the Ϫ1 kb region of eIF4E1 harbors six potential HREs and that there are two potential HREs in the proximal (ϽϪ100 bp) promoter region of eIF4E1 (Fig. 6A).
We performed ChIP assays with the four promoter regions of eIF4E1 and anti-HIF-1␣-specific antibodies. We observed that region 4 exhibited binding affinity to HIF-1␣, but regions 1, 2, and 3 did not (Fig. 6, B-E). Furthermore, we found that hypoxia treatment increased binding affinity between region 4 and HIF-1␣ in breast cancer cells, indicating that HIF-1␣ might bind the HREs in region 4 (Fig. 6F). To confirm the HREs in region 4 as the HIF-1␣ binding site, we generated the HRE mutant (ACGTG to AAAAG) and tested it with luciferase assays at normoxia and hypoxia. We observed that hypoxia increased luciferase activity of wild type, and overexpressed HIF-1␣ further elevated its activity, whereas point mutation abrogated these hypoxia-and HIF-1␣-promoted luciferase activities (Fig. 6G). We verified the above results in another type of breast cancer cell, MDA-MB-231 (data not shown). These results demonstrated that HIF-1␣ utilizes the HREs in region 4 of eIF4E1 to up-regulate eIF4E1 expression in breast cancer cells at low oxygen concentrations. Taken together, our data elucidate the previously unknown mechanisms of HIF-1␣ promotion of cap-dependent translation of selective mRNAs in hypoxic cancer cells.

DISCUSSION
In this study, not only did we provide evidence that HIF-1␣ promotes cap-dependent translation of selective mRNAs in hypoxic cancer cells, but more significantly, we unraveled the mechanisms by which HIF-1␣ activates eIF4E1. In particular, we elucidated that hypoxia-promoted tumorsphere growth of breast cancer cells is HIF-1␣-dependent. Hypoxia elevates expressions of eIF4E1 and eIF4E2, whereas HIF-1␣ preferentially promotes eIF4E1 expression in breast cancer cells. The expression of eIF4E3 is largely sequestered in breast cancer cells. We further demonstrated that HIF-1␣ promotes cap-dependent translation of a subset of mRNAs through up-regulating eIF4E1 (Fig. 7). We identified that HIF-1␣ utilizes HREs in the proximal promoter region of eIF4E1 to promote its transcription in hypoxic cancer cells. These findings provide new insights into the protein synthesis mechanisms of adaptive advantages in cancer cells in a low oxygen environment.
Importantly, our finding that HIF-1␣ promoted eIF4E1 expression in hypoxic cancer cells may explain the generally elevated eIF4E1 levels in a wide range of solid tumors, which associate with "chronic" hypoxia (when tumors outgrow their blood supply due to uncontrolled proliferation) and "acute" hypoxia (transient periods of low oxygen caused by aberrant blood flow). This study unravels the previously unknown roles of HIF-1␣ in translation via eIF4E and might supply a novel pathway of HIF-1␣/eIF4E1/eIF4F as targets for the treatment of hypoxic cancers. It is presumed that HIF-1␣-up-regulated eIF4E1 increases the availability of eIF4E1 for the translation of a subset of mRNAs with highly structured 5Ј-UTRs whose translation is highly eIF4E1-dependent.
Interestingly, during the submission of this study, Osborne et al. (65) reported that eIF4E3 competes with eIF4E1 in cap structure binding and therefore acts as a tumor suppressor. We found that the cellular eIF4E3 is very low compared with eIF4E1 in breast cancer cells at both normoxia and hypoxia, indicating that the cancer cells have acquired the capacity to sequester the tumor repression activity of eIF4E3 by unknown mechanisms.
Previous studies reported that HIF-2␣⅐RBM4⅐eIF4E2 complex-activated initiation is an alternative pathway for cap-de-pendent translation (18) and that eIF4E2 does not bind to eIF4G (38). We observed that eIF4E1 is most abundant among all eIF4E family members in breast cancer cells under both normoxia and hypoxia conditions, suggesting that eIF4E1, not eIF4E2, may be the primary regulator for cap-dependent translation of a subset of mRNAs, such as c-Myc, Cyclin-D1, and eIF4G1, in these breast cancer cells. The observations that hypoxic tumorspheres are more sensitive to the eIF4E1-eIF4G1 interaction inhibitor (E)-4EGI-1 compared with those at normoxia strongly suggest that eIF4E1-mediated cap-dependent translation, but likely not IRES-dependent or eIF4E2-dependent initiation, plays a primary role in translation initiation in hypoxic breast cancer cells. Our data indicate that agents targeting the eIF4E1-eIF4G1 interaction might be potent candidates for the treatment of cancer in a low O 2 environment.
The sensitivity of hypoxic tumorspheres to the eIF4E1-eIF4G1 interaction inhibitor (E)-4EGI-1 and the elevated expressions of eIF4E1, c-Myc, Cyclin-D1, and eIF4G1 suggest that hypoxia-suppressed mTOR activity does not affect the cap-dependent translation of selective mRNAs in cancer cells at hypoxia. This is in agreement with previous reports that breast cancer cells acquire resistance to uncoupling hypoxia-responsive signaling of mTOR activity from cap-dependent translation repression. Our observations that VEGF is dominantly regulated by eIF4E1-mediated cap-dependent translation are consistent with previous reports that VEGF is primarily mediated by IRES-independent translation (21). We observed that hypoxia promoted eIF4E2 expression, which was not significantly decreased by HIF-1␣ knockdown, suggesting that eIF4E2 The expression of eIF4E3 is largely inhibited in cancer cells at both normoxia and hypoxia. HIF-1␣ binds HREs in the eIF4E1 promoter region and up-regulates eIF4E1 expression. Increased eIF4E1 facilitates formation of eIF4E-eIF4G complex, elevates translation of selective mRNAs important for cancer cell adaptation to hypoxia stresses, and subsequently promotes cancer cell proliferation and tumorsphere growth at low O 2 conditions. Breast cancer cells acquire resistance to hypoxia by uncoupling the oxygen-responsive signaling pathway from mTOR function, eliminating suppression of protein synthesis mediated by hypophosphorylated 4E-BPs (1, 66) (in dashed frame). Black arrows or lines are from previous reports of other groups (1,66); blue arrows or lines are from this study. p-4E-BPs, hyperphosphorylation of 4E-BPs; AMPK, AMP-activated protein kinase; RHEB, Ras homolog enriched in brain. transcription is not (at least is not largely) regulated by HIF-1␣. The upstream pathway mediating eIF4E2 expression needs future elucidation.
The multidomain scaffold adaptor protein eIF4G1 is required for eIF4E1-involved cap-dependent and IRES-dependent translation initiation. We observed that hypoxia significantly elevated cellular eIF4G1, whereas HIF-1␣ depletion significantly decreased eIF4G1, suggesting that the expression of eIF4G1 in hypoxic cancer cells is at least partly HIF-1␣-dependent. Whether HIF-1␣-promoted eIF4G1 expression facilitates IRES-dependent initiation in hypoxic cancer cells is a concern for future work. In summary, our study demonstrated that HIF-1␣-promoted cap-dependent translation of a subset of mRNAs through up-regulating eIF4E1 is a driving element of tumor growth, providing a novel pathway of protein synthesis mechanisms in cancer cells at low oxygen levels.